whisker barrel cortex delta oscillations and gamma power in the

ARTICLE

Received 15 Oct 2013 | Accepted 6 Mar 2014 | Published 1 Apr 2014

Whisker barrel cortex delta oscillations and gammapower in the awake mouse are linked to respirationJ. Ito1,*, S. Roy2,*, Y. Liu2,*, Y. Cao2, M. Fletcher2, L. Lu2, J.D. Boughter2, S. Grun1,3 & D.H. Heck2

Current evidence suggests that delta oscillations (0.5–4 Hz) in the brain are generated by

intrinsic network mechanisms involving cortical and thalamic circuits. Here we report that

delta band oscillation in spike and local field potential (LFP) activity in the whisker barrel

cortex of awake mice is phase locked to respiration. Furthermore, LFP oscillations in the

gamma frequency band (30–80 Hz) are amplitude modulated in phase with the respiratory

rhythm. Removal of the olfactory bulb eliminates respiration-locked delta oscillations

and delta-gamma phase-amplitude coupling. Our findings thus suggest respiration-locked

olfactory bulb activity as a main driving force behind delta oscillations and gamma power

modulation in the whisker barrel cortex in the awake state.

DOI: 10.1038/ncomms4572 OPEN

1 Institute of Neuroscience and Medicine (INM-6) and Institute for Advanced Simulation (IAS-6), Julich Research Centre and JARA, Building 15.22, 52425Julich, Germany. 2 Department of Anatomy and Neurobiology, University of Tennessee Health Science Center, 855 Monroe Ave, Memphis, Tennessee 38163,USA. 3 Theoretical Systems Neurobiology, Institute for Biology II, RWTH Aachen University, 52056 Aachen, Germany. * These authors contributed equally tothis work. Correspondence and requests for materials should be addressed to D.H.H. (email: [email protected]).

NATURE COMMUNICATIONS | 5:3572 | DOI: 10.1038/ncomms4572 | www.nature.com/naturecommunications 1

& 2014 Macmillan Publishers Limited. All rights reserved.

mailto:[email protected]

http://www.nature.com/naturecommunications

Oscillatory neuronal activity is a common phenomenon inmammalian neocortex and can be readily observed inlocal field potential (LFP) and electroencephalogram

(EEG) signals1. Oscillations occur across a broad rangeof frequencies including slow (0.5–8 Hz) delta/theta and high-frequency gamma band (30–80 Hz) oscillations2. How oscillatorybrain activity is generated and controlled, particularly in the low(delta, 0.5–4 Hz)-frequency range, and how oscillations relate tobehaviour are only partially understood3. Neocortical neuronaloscillations in the delta frequency range (0.5–4 Hz) are believed tobe generated by mechanisms involving both neocortical andthalamic neuronal circuits4,5 and are commonly associated withslow-wave sleep in both humans and animals. However, delta

band oscillatory activity in LFP and spike activities was alsoobserved in the whisker barrel cortex of awake, head-fixed mice6–8.At rest, mice breathe rhythmically at frequencies between 1 and3 Hz, which corresponds to the delta frequency range of neuronaloscillations. While it has been shown on a behavioural level thatrespiration and whisker movements are dynamically phase lockedin this frequency band9–11, whether the neuronal rhythm in thebarrel cortex is related to respiration has not been investigated.

To examine a possible link between barrel cortical deltaoscillations in awake mice breathing at resting respiratory rate, wesimultaneously measured breathing rhythm and neuronal activityin the whisker barrel cortex of awake head-fixed mice. We foundthat delta band neuronal oscillations in the whisker barrel cortex

2

–2

Res

pira

tion

Res

pira

tion

LEP

s (m

V)

LEP

s (m

V)

Nor

mal

ized

cor

rela

tion

Nor

mal

ized

cor

rela

tion

217 218 219 220

181 182 183 184Lag (s)

Lag (s)

Auto(150 – 200 s)

Auto(200 – 250 s)

Resp-LFP cross(150 – 200 s)


Lag (s)

Lag (s)

Time (s)

Time (s)

Hypoxia Normoxia

Time (s)

Resp-LF

P coherence

0.0

0.2

0.4

0.6

0.8

1.0

100 150 200 300250

–1

0

1–1

0

1

–1

0

1

–1

0

1

–1 0 1 –1 0 1

–1 0 1 –1 0 1

0

2

4

2

0

0

Freq

uenc

y (H

z)

4

–1

0

1

2

–2

–1

0

1

0

Figure 1 | Oscillatory neuronal activity phase locks with breathing. Correlation and coherence between respiratory and local field potential (LFP) signals

in barrel cortex in an awake mouse. Data from one mouse are shown, but results were reproduced in five out of five mice. (a) Respiratory rhythm

(black) and two LFP signals simultaneously recorded in the whisker barrel cortex (red and blue) at rest. Recording sites were 300mm apart.

(b) Autocorrelations (left) of respiratory and LFP signals and LFP-respiration cross-correlations (right). Curve colours correspond to those in a.

(c) Same recordings as in a, but during accelerated breathing induced by brief exposure to hypoxic air. (d) Autocorrelations (left) of respiratory and

LFP signals and LFP-respiration cross-correlations (right). Curve colours correspond to those in a. (e) Instantaneous respiration frequency (top) and

LFP-respiration coherency (bottom) during normal accelerated breathing elicited by exposure to hypoxic air. Respiration frequency and coherency

were computed within 5-s time windows, moved in 2.5-s steps. The coherence was plotted in a pseudo-colour code shown on the right. Grey

background in top panel indicates time of exposure to hypoxic air. Data shown in a and c are two 3-s segments of the full data range shown in e

(the time axes in a and c are common with the one in e).

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4572

2 NATURE COMMUNICATIONS | 5:3572 | DOI: 10.1038/ncomms4572 | www.nature.com/naturecommunications



were phase locked to the respiratory rhythm and were mostlydriven by olfactory bulb activity. Furthermore, phase–amplitudecoupling analysis showed that LFP oscillations in the gammafrequency band were amplitude modulated in phase with therespiratory rhythm. Thus, our results demonstrate that respira-tory activity directly modulates slow (1–4 Hz), rhythmic neuronalactivity in the somatosensory whisker barrel cortex and indirectlymodulates gamma band power through phase–amplitude cou-pling mechanisms. Several studies have linked theta-gammaphase–amplitude coupling to sensory, motor and cognitiveprocesses12–15. Our findings thus suggest a possible neuronalmechanism for cross-modal interactions between olfaction andwhisker tactile sensation9–11.

ResultsRespiration-locked LFP oscillations in the barrel cortex. Usingmultiple-electrode extracellular recording techniques, we mea-sured spike and local field potential (LFP) activity in the soma-tosensory whisker barrel cortex of awake, head-fixed mice.Simultaneously, respiratory activity was monitored non-invasivelyby measuring the air temperature fluctuations in front of thenostrils (see Methods). LFP as well as single- and multi-unit spikeactivity were rhythmically correlated with respiration (Figs 1,2).

Effects of olfactory bulbectomy on oscillations. Respiration-locked neuronal activity is known to occur in the rodent olfactorybulb, driven by the activation of olfactory sensory neurons witheach breath through the nose16,17. Fontanini et al.18,19 haveshown that these oscillations can propagate to the olfactorypiriform cortex. The average respiratory rate in C57BL/6J mice atrest is 165 breaths min� 1 or 2.75 Hz (ref. 20). Hence, micebreathe at a resting rate that lies within the 1–4 Hz deltafrequency range defined for neuronal oscillations1. We askedwhether respiration-locked neuronal delta band oscillations in thesensory whisker barrel could be driven by olfactory bulb activity.As a first experimental approach, we surgically removed theolfactory bulb (in n¼ 6) mice. Olfactory bulbectomies eliminatedbarrel cortical delta band oscillations and the correlation betweenrespiration and LFPs (Fig. 3). We quantified this reduction in thelocking between respiration and LFP in terms of their coherenceat the respiration frequency. This decrease in the LFP-respirationcoherence due to bulbectomy corresponds to an averagereduction by 91% (Fig. 3), hence a highly significant decrease(Po0.001 (two-sides), Mann–Whitney U-test).

LFP oscillations related to nasal airflow. To control if othereffects not related to respiration were responsible for the loss ofcoherence between respiration and LFP in bulbectomized mice, weperformed tracheotomies in intact, anaesthetized mice (n¼ 3) toeliminate respiration through the nose. This manipulation selec-tively prevented activation of the olfactory bulb through nasalairflow while leaving the bulb intact. LFP activity was recorded inthe whisker barrel cortex and respiratory activity was monitoredthrough chest movements. In addition, we generated an artificialrhythmic airflow through the nose independent from the sponta-neous tracheal respiration (Fig. 4). This was performed at threedifferent frequencies of airflow and a controlled volume thatmimicked the mouse’s resting tidal volume of B0.15 ml (seeMethod). Frequencies for artificial nasal airflow were chosen to belower (0.8 Hz), higher (3.2 Hz) or the same (1.6 Hz) as the averagespontaneous respiratory rate of the anaesthetized mouse. In tra-cheotomized mice, the correlation between respiration and barrelcortical LFP activity was weak (Fig. 4b,d,f, right, green), but theLFP activity was strongly correlated with the rhythmic nasalairflow (Fig. 4b,d,f, right, black) at all three frequencies tested.

The highest correlation between nasal airflow and LFP activityoccurred at a 1.6-Hz rhythm of nasal airflow. This could be due toresonance phenomena in cortical principal neurons, which havebeen shown to possess a preferred subthreshold frequency close to1.6 Hz (ref. 21).

LFP activity follows electrical olfactory bulb stimulation. Therhythmic airflow through the nose might exert small mechanicalforces on the olfactory bulb that could cause neuronal responsesdue to small movements of the brain tissue against the recordingelectrodes. We therefore performed another control experimentin which we eliminated nasal airflow by performing a tracheot-omy, but now modulated neuronal activity in the olfactory bulbthrough electrical stimulation while recording barrel cortical LFPactivity. We used rhythmic electrical stimulation consisting ofbrief trains of current steps to activate the ipsilateral olfactorybulb at the same frequencies used for the artificial nasal airflowstimulation (0.8, 1.6 and 3.2 Hz) in the previous experiment.These stimuli caused stimulus-locked rhythmic fluctuations of thebarrel cortical LFPs (Fig. 5), confirming that rhythmic olfactorybulb activity rhythmically modulates neuronal activity in thewhisker barrel cortex.

Inte

rval

cou

nt

30

20

10

0

–0.005

0.0

0.005

0.010

0.015

Cor

rela

tion

coef

.–1.0 –0.5 0.0 0.5 1.0

Interval duration (s)

Lag time (s)

1 ms

0.2

mV

0 0.1 0.2 0.3 0.4 0.5

Figure 2 | Spike activity is rhythmically correlated with respiration.

Single- (n¼ 3) and multi-unit (n¼9) spike activity was recorded in the

whisker barrel cortex of two awake mice. The cross-correlations of spike

activity with breathing was significant for 3/3 single units and 8/9 multi-

unit recordings. (a) Inter-spike interval histogram of single-unit spike

activity recorded over 10 min in the whisker barrel cortex of an awake

mouse. Inset shows overlays of waveforms of the first (left) and last (right)

10 action potentials. (b) Cross-correlation of single-unit spike activity with

respiration (time of onset of the inspiratory cycle). The blue line represents

the raw cross-correlation. The black line represents the median of the

distribution of surrogate correlations. Top and bottom red lines represent

the 5th and 95th percentile of the surrogate correlation distribution,

respectively (see Methods or details). Raw correlations were considered

significant if cross-correlation values exceeded the 5 or 95 percentile

boundaries of the surrogate correlation distributions.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4572 ARTICLE




Phase–amplitude coupling of gamma and respiratory rhythms.Recent studies suggest that gamma band amplitude or power ismodulated in phase with delta band oscillations, a finding referredto as phase–amplitude coupling14,15,22. Our data also show this typeof cross-frequency interaction. We found that the power of gammaoscillations (75 Hz) in the whisker barrel cortex was modulated inphase with the respiratory cycle (Fig. 6). Thus, not only slowoscillations but also higher-frequency activities in the whisker barrelcortex are linked to respiration. Such cross-frequency coupling hasbeen implicated in spatial and temporal coordination of neuronalactivity associated with higher brain functions14,23. By use of abootstrap statistical approach, using phase-randomized surrogatedata (see Methods) we show that removal of the olfactory bulbeliminated the respiration-related modulation of power around thefrequency range of transition from gamma (30–80 Hz) to fast(80–200 Hz) oscillations. The opposite occurred in frequency rangesbelow and above this band where phase–amplitude coupling wasobserved after olfactory bulbectomy (Fig. 6).

Retrograde tracing. The pathway through which respiration-locked neuronal activity reaches the whisker barrel cortex is

currently unknown. The large olfactory or piriform cortex in miceis known to show respiration-locked neuronal oscillations18,19

and is thus a plausible link in the pathway from the mainolfactory bulb to the barrel cortex. To determine whether thewhisker barrel cortex receives direct projections from the piriformcortex or other olfaction-related areas, such as the entorhinalcortex, the amygdala or the orbitofrontal cortex24,25, we injected aretrograde tracer (Fluorogold) into the barrel cortex (seeMethods). Labelling of forebrain neurons projecting to thewhisker barrel cortex was consistent with previous studies(for example, ref. 26). Projection neurons were found in ventralposteromedial thalamus (VPM) and to a lesser extent as well inventral posterolateral thalamus (VPL). Labelled neurons werealso found in several cortical areas, including contralateralbarrel cortex, as well as adjacent ipsilateral cortical areas,including secondary somatosensory (S2), entorhinal andinsular cortex and rostral motor cortex. In somatosensorycortical areas, projection neurons were predominantly found inlayers 2,3 and 5, but not in layers 1 or 4. Sparse, large labelledneurons were found in basal forebrain nuclei—these are likelycholinergic neurons in the nucleus basalis known to project tomultiple cortical areas.

Res

pira

tion

LEP

s (m

V)

–1

0

1–1

0

1

2

–2

0

Nor

mal

ized

cor

rela

tion

Lag (s)110 111 112 113

Lag (s)

0

1

–11–1

Time (s)

0 –1 0 1

Auto(100 – 150 s)


Res

p-LF

P c

oher

ence

0 1 2 3 40.0

0.2

0.4

0.6

0.8

1.0

Respiration frequency (Hz)

Res

p-LF

P c

oher

ence

0.0

0.2

0.4

0.6

0.8

1.0

Normal Bulbecto.

Animal type

Figure 3 | Olfactory bulb activity drives cortical oscillations. Reduction of coherence between respiration and local field potential (LFP) signals in intact

and olfactory bulbectomized mice recorded in awake, head-fixed conditions. (a) Respiration (black) measured as the temperature in front of the

mouse’s nostrils and simultaneously recorded LFP signals (red and blue) in the barrel cortex of a bulbectomized mouse. The respiration signal was

normalized by subtraction of the mean and division by the s.d. (b) Autocorrelations (left) of the LFP and respiratory signals shown in a and cross-

correlations (right) between the respiratory signal and either of the LFP signals in a. Red and blue curves represent the cross-correlations obtained from the

LFPs drawn by the corresponding colour in a. (c) Coherence between respiratory signals and LFP signals plotted against respiration frequency. Individual

data points represent the coherence within a data segment of 50 s between respiration and the simultaneously recorded LFP signal. Different symbols

represent different mice, and white and black symbols represent intact and bulbectomized mice, respectively. A total of 52 and 18 50-s segments

were sampled from five intact and six bulbectomized mice, respectively. Multiple identical symbols represent multiple parallel LFP measurements

(with up to five electrodes in the whisker barrel cortex) or multiple LFP samples from the same recording site in one mouse. Each symbols position along

the x axis is determined by the average respiratory rate during the corresponding 50-s segment. (d) Summary of the data shown in c. White and

black box-and-whisker plots represent the means (ticks inside the boxes), the 1st and 3rd quartiles (bottom and top ends of the boxes) and the extents of

the data (whiskers) for the white and black data points in c, respectively.





DiscussionWe find that respiration-locked olfactory bulb activity caused byairflow through the nose is a major driving force behind deltaband neuronal oscillations in the somatosensory cortex of awakemice. Coherence of cortical neuronal oscillations with respirationwas reduced by 91% after removal of the olfactory bulb (Fig. 3).Somatosensory cortical oscillations could be directly linked toolfactory bulb activity by experiments involving controlledairflow activation of the bulb in tracheotomized mice (Fig. 4)or through electrical stimulation of the bulb (Fig. 5). Both

manipulations result in the modulation of barrel cortical activityphase locked to the airflow or electrical stimulation.

Adrian27 showed already in 1950 that respiration drivesoscillatory spike and LFP activity in the mammalian olfactorybulb. A recent study in rats confirmed that respiration-lockedoscillations in the bulb depend on rhythmic airflow throughthe nose17. These oscillations do not require the detection ofodours because mammalian olfactory sensory neurons are alsomechanical sensors, signalling pressure changes caused by nasalairflow28.

Nas

al a

irLF

P (

mV

)C

hest

–0.2

0.2

0

–1

0

1

1

–1

0

Nas

al a

irLF

P (

mV

)C

hest

148 149 150 151

80 81 82 83

–0.2

0.2

0

–1

0

1

1

–1

0

Nas

al a

irLF

P (

mV

)C

hest

–0.2

0.2

0

–1

0

1

1

–1

0

Lag (s)330 331 332 333

Time (s)

Time (s)

Time (s)

Lag (s)1–1 0 –1 0 1

Nor

mal

ized

cor

rela

tion

Auto (330 – 380 s) Cross (330 – 380 s)

0

1

–1

Lag (s) Lag (s)1–1 0 –1 0 1

Lag (s) Lag (s)1–1 0 –1 0 1

Nor

mal

ized

cor

rela

tion

Auto (50 – 100 s)

Auto (120 – 170 s)

Cross (50 – 100 s)

Cross (120 – 170 s)

0

1

–1N

orm

aliz

ed c

orre

latio

n

0

1

–1

Figure 4 | Nasal airflow is required to link respiration to cortical oscillations. Local field potential (LFP) oscillations in whisker barrel cortex related to

nasal airflow in an anaesthetized tracheotomized mouse. Data are shown from one mouse, but results were reproduced in 3/3 mice. (a) Raw signals

representing nasal airflow modulated at 0.8 Hz through the opening of the ascending trachea (black), spontaneous breathing through the opening

of the descending trachea measured as chest movement (green) and barrel cortical LFP (light red: raw signal, red: filtered in 0.5–10 Hz).

(b) Autocorrelations (left) of the signals in a with the same colour convention, and cross-correlations (right) of the filtered LFP signal to the nasal airflow

(black) and to the chest movement (green). All the correlations were computed from data segments of 50 s duration that include the traces shown in a, as

indicated by the labels within the panels. (c) As in a with nasal airflow modulated at 1.6 Hz. (d) Auto- and cross-correlations as in b for the

signals in c. (e) As in a with nasal airflow modulated at 3.2 Hz. (f) Auto- and cross-correlations as in b for the signals in e. The units of the nasal

airflow and the chest movement are arbitrary, but common across a,c and e.





LFP

mod

ulat

ion

rela

tive

to m

axim

um

1.0

0.9

0.8

0.7

0.6

0.5

0.4

Bulbar stimulation rhythm0.8 Hz 1.6 Hz 3.2 Hz

80 µA50 µA

3.2 Hz, 80 µA

0.8 Hz, 80 µA0.8 Hz, 50 µA

Stim.artif.

1.6 Hz, 80µA

100

µV

1 s

Figure 5 | Electrical stimulation of the olfactory bulb entrains cortical activity. (a) Average barrel cortical LFP response to electrical stimulation of the

ipsilateral olfactory bulb with a 0.8-Hz rhythmic stimulus. The stimulus consisted of 50 Hz trains of 10-ms current steps (50mA) repeated every 1.25 s

(¼0.8 Hz). The arrow points to the onset of the stimulus train. Electrical stimulation artefacts (marked by curly brackets labelled stim.artif.) appear as

dense vertical lines truncated at the bottom. LFP signals were averaged across 30 stimulus trains. The LFP increased during bulbar stimulation and

decreased again during the inter-train interval. The average LFP response amplitude was measured as the difference between the voltage at the time of the

onset and end of the stimulus train in the averaged signal, as marked by the two dashed horizontal lines. (b) Average barrel cortical LFP response to the

same stimulus as in a but with a higher stimulus current (80mA). The LFP increased faster with higher stimulus current. (c,d) Average barrel cortical LFP

response to bulbar stimulation (80mA) at 1.6 Hz (c) and 3.2 Hz (d). In all cases, bulbar stimulation was through a concentric bipolar electrode (SNEX 100x,

Kopf Instruments, USA). Stimulus timing was controlled using Spike2 software and the D/A output of the CED 1401 (both Cambridge Electronic Design,

UK) and stimulus currents were generated by a battery driven stimulus isolation unit (A380, World Precision Instruments, USA). (e) Average LFP

responses measured as described in a to three different stimulus frequencies at two different stimulus currents. Responses to each stimulus frequency-

current combination were measured in three mice. Response amplitudes are normalized to each animal’s maximum response to the 0.8 Hz/80mA stimulus

and averaged across the three mice. Response amplitude was a function of stimulus current and stimulus frequency. Error bars show s.e.m.

Delta(1.5 – 4 Hz)

Delta(1.5 – 4 Hz)

Theta(4 – 10 Hz)

Theta(4 – 10 Hz)

Beta(10 – 30 Hz)

Beta(10 – 30 Hz)

Gamma(30 – 80 Hz)

Gamma(30 – 80 Hz)

Fast(80 – 200 Hz)

Fast(80 – 200 Hz)

Normal (200 – 250 s)Respiration: 3.4 Hz

Bulbectomized (10 – 60 s)Respiration: 2.9 Hz

Samples from normal (N= 52)

Samples frombulbectomized (N= 18)

Mea

n ve

ctor

leng

th

Frequency of amplitude (Hz)

Frequency of amplitude (Hz)

1 2 4 8 16 32 64 128 256

1 2 4 8 16 32 64 128 2560.0

0.1

0.2

0.3

0.4

0.5

Frac

tion

of s

ign.

sam

ples

0.08

0.04

0.00

0.04

0.00

0.08–3

Time (s)42 43 44 45

240 241 242 243Time (s)

–3

3

0

�-ph

ase

�-am

p.

Normal (200 – 250 s)�: 3.4 Hz, �: 75 Hz

Bulbectomized (10 – 60 s)�: 2.9 Hz, �: 75 Hz

�-phase–3 –2 –1 0 1 2 3

�-phase–3 –2 –1 0 1 2 3

1.2

1.0

0.8

�-am

plitu

de

1.2

1.0

0.8

�-am

plitu

de

Res

pira

tion 1

0

–13

0

–3

3

0

�-ph

ase

�-am

p.R

espi

ratio

n 1

0

–1

–3

3

0

Figure 6 | Respiratory modulation of power in high-frequency oscillations. Phase–amplitude coupling between respiration-locked delta and gamma band

oscillations in the barrel cortical LFP activity of an awake intact and an awake bulbectomized mouse, followed by population statistics. (a) Respiratory

activity (top trace), amplitude of gamma band oscillations (middle trace) and delta oscillations (light green bottom trace) and its phase (dark green

bottom trace) in an intact mouse. Gamma oscillation (75 Hz) amplitude peaks rhythmically phase locked to the delta cycle. (b) Gamma oscillation

amplitude as a function of delta phase (red). The solid and dotted black lines indicate the mean and the 2.5 and 97.5 percentile boundaries of the surrogate

amplitude distribution estimated from 1000 phase-randomized surrogates. Gamma amplitude modulation is significant at phase 0 of the delta cycle.

(c,d) Same as a,b, respectively, but for a bulbectomized mouse. After removal of the olfactory bulb, the amplitude modulation of the gamma band

oscillations is no longer phase locked to respiration. (e) Quantitative analysis of phase–amplitude coupling by the mean vector length (MVL, solid black

lines) between the phase of the respiratory (delta) frequency and the amplitudes of a range of frequencies from 1 to 256 Hz (logarithmically scaled). Grey

and white backgrounds delineate commonly used frequency band boundaries1. The solid and dotted grey lines indicate the mean and the 95 percentile,

respectively, of surrogate mean vector length values estimated from 1,000 phase-randomized surrogates. (f) Population statistics of respiration-LFP

phase–amplitude coupling in intact and bulbectomized mice. MVL values were computed from the same set of LFP segments as used in the coherence

analysis in Fig. 2c (52 samples from five intact and 18 samples from six bulbectomized mice). The graphs represent the fraction of samples exceeding the

95 percentile boundary of the surrogate MVL value distribution in intact (black) and bulbectomized mice (grey) plotted as a function of amplitude

frequency (logarithmically scaled). Phase frequency was fixed at the respiratory frequency.





The influence of respiration-locked olfactory bulb activity onthe whisker barrel cortex is not limited to delta band oscillations.Phase–amplitude coupling analysis reveals that the power ofgamma band oscillatory activity in the whisker barrel cortex ismodulated in phase with respiration-locked delta band activity.

An interaction between respiration as a behaviour linked toolfaction and active whisker movements for tactile explorationhas been suggested by Welker, based on early behaviouralobservations in rats29. Recent studies confirmed that respiratoryand whisker tactile behaviours are coordinated in mice9 andrats10, suggesting a coordinated interaction of olfactory andtactile sensory acquisition. The fact that the two sensory systemsare not independent is further supported by studies showing thatthe loss of one results in increased sensitivity in the other sensorymodality30,31.

Whisking and breathing movements in mice are dynamicallycorrelated with correlations being particularly strong during slowwhisking (o5 Hz)9. Thus, during slow whisking, sensory inputfrom the whiskers is likely to contribute to the breathing-LFPcoherence in the whisker barrel cortex, in addition to olfactorybulb sensory input. However, the contribution of whisker sensoryfeedback to breathing-LFP coherence can be no larger thanB10% of the maximal coherence, since bulbectomy experimentsshow that olfactory bulb activity accounts for B90% ofLFP-breathing coherence.

Since delta band oscillations in cortical activity are also found inmammals who breathe much slower than mice (for example,humans), parallel neuronal mechanisms likely involving thalamo-cortical circuits exist that generate delta oscillations independentof the respiratory rhythm. Our findings suggest, however, thatrespiration might cause a rhythmic modulation of somatosensorycortical activity at the species-specific respiratory rates. Removingthe olfactory bulb reduces the breathing-LFP coherence at therespiratory frequency by 91%. Thus, in mice olfactory bulb activitywas responsible for most of the respiration-locked neuronaloscillations, but other respiration related to sensory inputs, e.g.,from the trigeminal nasal airflow sensors, upper airway sensorsand chest muscles, are likely to contribute as well. Those could beresponsible for the residual respiration-locked modulation

observed in bulbectomized mice and would also be present inanimals with poor or no sense of olfaction, like dolphins.

Our findings provide a potential neuronal mechanism for theinteraction between olfactory and whisker tactile sensation inawake-behaving mice. Behavioural studies have shown that thetiming of respiratory and whisker movements are dynamicallycoordinated9,11. The results by Cao et al.9 suggest that activeodour and tactile sensing through sniffing and high-frequencywhisking can occur independently in mice , while a study byDeschenes and colleagues suggests continuous strong phaselocking between the two rhythms in rats11.

Gamma oscillations in the barrel cortex of rats have beenshown to increase immediately before exploratory whiskingbehaviour32 and after mechanical whisker stimulation33. Whetherthese changes in gamma activity were linked to respiration wasnot addressed in those studies. Our results suggest that themodulation of barrel cortical gamma band power in phasewith respiration could play a role in coordinating olfactory andtactile sensory processing. Additional clues about the potentialfunctional role of respiration-locked oscillations in whiskersensory processing might come from a comparative investi-gation of different species. Rats, gerbils, hamsters and chinchillas,who also have barrel cortices and motile whiskers, also haveresting respiratory rates within the delta frequency band (B1.2,1.4, 0.7 and 0.7 Hz, respectively). On the basis of our findingsthose species could be expected to show respiration-lockedgamma power modulation in the barrel cortex, similar to mice. Incontrast, Guniea pigs possess a whisker barrel cortex but haveimmotile mystacial whiskers and thus lack active whisker-sensingbehaviour. If respiration-locked modulations of gamma powerdoes play a role in the coordination of respiration with whiskertactile sensing, it could be less pronounced or even absent inanimals without active tactile sensing.

In summary, respiration-locked neuronal oscillation in theolfactory bulb is a major driving force behind delta band neuronaloscillations in the somatosensory cortex of mice. Furthermore,gamma band oscillations, which have been widely implicated inhigher cortical processing, are amplitude modulated in phasewith respiration. Thus, the life supporting respiratory behaviour,

S1BF

S2

IC

VPL

VPMVL

1

3

2

4

5

BF

LH

Contra S1BF

65

43

21

S2

PC

Figure 7 | Retrograde tracing of barrel cortical afferents. Representative images showing retrograde labelling in the forebrain following injection

of a retrograde tracer (Fluorogold) into barrel cortex. (a) Injection site in S1BF (barrel cortex). (b) Retrogradely labelled cells were not found in the

piriform cortex (PC). Retrogradely labelled neurons are found primarily in layers 2,3 and 5 in the contralateral barrel cortex (c), ipsilateral ventral

posteromedial (VPM) thalamus (d), adjacent ipsilateral S2 (somatosensory) cortex (e). There is also sparse labelling of neurons ipsilaterally in basal

forebrain (BF) nuclei (f). IC, internal capsule; LH, lateral hypothalamus; VPL, ventral posterolateral thalamus; VL, ventral lateral thalamus. Scale bars,

(a), 400 mm. (b–f), 100mm. Images a–c and d–f are from different mice.





believed to be controlled mostly by brain stem centres, has anunexpected, indirect influence on higher-order brain activity inmice, via the olfactory system. Whether similar influences ofbreathing on brain activity also exist in other cortical areas andpossibly also in humans, whose olfactory bulb is much reduced insize relative to the neocortex, remain to be shown.

MethodsAnimals. Experiments were performed on adult male C57BL/6J (B6) mice (48weeks old, 18–25 g body weight). Male mice were used to avoid variability inbehavioural physiological measures related to oestrous cycle. Mice were housed in abreeding colony with 12-hour light/dark cycles in standard cages housing maxi-mally five adult mice with ad libitum access to food and water. All experimentswere performed during the light cycle (between 1200 and 1700 hours). None of themice had undergone any previous experimental procedure. All animal experi-mental procedures adhered to guidelines approved by the University of TennesseeHealth Science Center Animal Care and Use Committee. Principles of laboratoryanimal care (NIH publication No. 86–23, rev. 1996) were followed.

Preparation for awake, head-fixed recording. For surgeries, mice were anaes-thetized with 3% Isoflurane (Baxter Pharmaceutical Products, Deerfield IL) inoxygen in an incubation chamber, transferred to a stereotaxic head mount andanaesthesia was continued with 1–2.5% Isoflurane in oxygen through a mouth-piece. Isoflurane concentration was controlled with a vaporizer (Highland MedicalEquipment, CA). The depth of anaesthesia was adjusted until the mice failed toshow a reflex withdrawal of the hind paw to a strong pinch. Blunt ear bars wereused to prevent damaging the eardrums. Core body temperature, measured with arectal thermometer, was maintained between 36.5 and 38.0 �C with a feedback-controlled heating pad (FHC Inc., Bowdoinham, ME). Surgical techniques weredescribed in detail elsewhere34,35. In brief, a small craniotomy (diameter, 1-2 mm)was made over the right whisker barrel cortex 1 mm posterior to Bregma and3.5 mm lateral to the midline. The exposed but intact dura was covered with TripleAntibiotic (Walgreens, US) to keep it moist and reduce the risk of infection. Acylindrical plastic chamber (0.45 cm diameter and 8 mm height) was placed overthe skull opening and filled with Triple Antibiotic. Three small machine screws(1/80 dome head, 0.8 mm diameter, 2 mm long, Small Parts, Inc., Miami Lakes, FL)were secured in the skull bone, and metal head post was mounted anterior toBregma. The chamber, head post and skull screws were secured into place withdental acrylic, as described earlier. Mice were injected subcutaneously with5 mg kg� 1 analgesic Torbugestic (Fort Doege, USA) to alleviate pain and 0.5 ml oflactated ringer solution as a fluid supplement twice within the first 24 h of thesurgery.

Electrophysiological recordings and respiration monitoring in awake mice.Mice were allowed a 3–4-day recovery period after surgical mounting of a headpost and recording chamber before being adapted to the head-restrained experi-mental situation during two sessions of head fixation of 15 min duration performedon the same day at 0900 and 1500 hours. During these sessions, the head was heldfixed and the body was covered with a loose fitting plastic half-tube (5 cm diameter,10 cm long) to limit movements. Mice typically adapted to the head fixation within2–3 sessions as judged by markedly reduced walking and running movementsduring the third session compared with the first or second. The experimental setup,head-holding device and recording procedures have been described in detail in atechnical publication34. In short, head fixation involved clamping the metal headpost mounted on the mouse’s skull to a metal fixation device, using a machinescrew. Triple antibiotic paste was then removed from the recording chamber, andthe chamber was rinsed and filled with sterile saline solution.

Before each experiment, mice were allowed to accommodate to the headfixation situation for 30 min before recordings.

For extracellular recordings, the guiding tubes of a computer-controlledmicrodrive (MiniMatrix, Thomas Recording, Germany) were lowered into thesaline-filled recording chamber to a distance of less than 2 mm from the duralsurface of the brain. The stainless steel guiding tubes also serve as referenceelectrodes and are eclectically connected to the brain tissue via the saline solution.Then, 2–5 electrodes (glass insulated tungsten/platinum, impedance: 3.5–5.0 MO)were slowly advanced through the intact dura into the whisker barrel cortex.Electrode movements were controlled with micrometre resolution and digitallymonitored. Local field potentials and spike signals were separated by bandpassfiltering at 0.1 to 200 Hz and at 200 Hz to 8 kHz, respectively, using a hardwarefilter amplifier (FA32; Multi Channel Systems). Filtered and amplified voltagesignals were digitized and stored on a computer hard disk (16 bit A/D converter;sampling rate, 420 kHz for action potentials, 41 kHz for LFPs) using a CEDpower1401 and Spike2 software (both Cambridge Electronic Design).

Respiratory behaviour was monitored based on temperature changes associatedwith the expiration of warm air. A thermistor (Measurement Specialties, Boston,MA, USA) was placed in front of one nostril (or descending trachea in sometracheotomy experiments), and breathing cycles could reliably be measured astemperature increased and decreased during exhale and inhale movements,

respectively. Increased temperature is represented as positive deflections of thethermistor voltage signal, which thus correspond to exhale movements (Figs 1,2).The raw thermistor voltage signal was digitized at 1 kHz and stored together withthe electrophysiological signals.

On completion of each recording session, the Ringer’s solution was removed,the recording chamber filled with triple antibiotic and the mice were returned totheir home cages. Each animal typically participated in experiments for 1–2 weeks.

Tracheotomy and nasal airflow control. To further investigate the questionwhether neuronal activity was locked to respiration as a motor act or to nasalairflow and hence olfactory bulb activation, we opened the trachea in anaesthetizedmice to separate respiration from nasal airflow. While the mouse could breathespontaneously through the descending trachea, we were able to independentlymodulate nasal airflow by moving air through the ascending trachea.

For the performance of a tracheotomy before electrophysiological recordings,mice (n¼ 3) were anaesthetized with Avertin (400 mg kg� 1). After completion ofthe tracheotomy, anaesthesia was continued with 1–2% Isoflurane in oxygeninhaled through the opening of the descending trachea. Spontaneous respirationwas monitored using a piezo film touching the chest wall. The ascending tracheawas connected to a Picospritzer III (Parker Hannifin Corp., Cleveland, OH, USA)to modulate nasal airflow by passing air (mimicking a B0.15 ml tidal volume) intothe trachea. Nasal airflow was modulated rhythmically to be half, the same anddouble the spontaneous respiration frequency (1.6 Hz). Nasal airflow wasmodulated at 0.8, 1.6 or 3.2 Hz and was monitored using a thermistor placed infront of the nostril as described in detail below.

Electrical stimulation of olfactory bulb. To investigate whether local fieldpotential activity in the whisker barrel cortex could be modulated by electricalstimulation of the olfactory bulb in the absence of nasal airflow, we performedexperiments in anaesthetized, tracheomotized mice (n¼ 3). A tracheotomy wasperformed under Avertin anaesthesia to eliminate nasal airflow and thus respira-tion-related bulbar activity. Anaesthesia was then continued using Isofluraneinhaled through the pulmonary trachea. Mice were mounted in a stereotaxicframe, and the skull was opened above the right whisker barrel cortex and olfactorybulb. Recording electrodes were inserted into the whisker barrel cortex. A con-centric bipolar stimulation electrode was lowered onto the intact dura of theolfactory bulb. We electrically stimulated the bulb with rhythmic stimulationpatterns consisting of 50 Hz trains of stimuli with inter-train intervals of 1.25, 0.62and 0.28 s, corresponding to stimulation rhythms of 0.8, 1.6 and 3.2 Hz,respectively. Stimulus trains consisted of 10-ms current steps using 50 or 80 mAstimulus currents. We kept the train frequency and step duration constant whilechanging the number of pulses to generate different stimulation frequencies.The stimulus trains had 30, 15 and 8 current steps for 0.8, 1.6 and 3.2 Hz stimu-lation rhythms, respectively.

Creation of hypoxic conditions. We increased the frequency of respiratoryactivity by exposing the mice for 1 min to hypoxic (B10% oxygen) air whilecontinuously recording the local field potentials. To this end, a custom-designedplastic cover was placed over the mouse leaving access to the recording chamberand head post through an opening at the top. Respiratory behaviour was monitoredwith a thermistor as described above. The composition of the air in the chamberwas controlled by mixing atmospheric air with nitrogen at a ratio of 1/0 fornormoxic and 1/2 for hypoxic conditions. Gas flowed into the chamber at a rate of1 l min� 1. An O2 analyser (GB-300, Teledyne Anal. Instr., City of Industry, CA,USA) continuously monitored the O2 concentration in the chamber. Following atleast 5 min of stable LFP recording, mice were exposed to hypoxic air (10% O2) for1 min, followed by normoxic air. Respiratory frequency, blood O2 saturation andheart rate were measured using a MouseOx (Starr Life Sci.Corp. Pittsburgh, USA)and recorded together with electrophysiological data before, during and afterexposure to hypoxic air.

Analysis of local field potentials. LFP data were collected in awake, head-fixedconditions from five intact and six olfactory bulbectomized mice. Primary analysisinvolved identification of the correlation of the LFP activity to the respiratoryactivity and the frequency of the resting respiratory rhythm. Periods of stableresting respiratory rhythm (o5 Hz) were selected for further analysis. Coherenceanalysis focused on the resting respiratory frequency range (1–3 Hz) only,excluding higher respiratory rates associated with the brief episodes of sniffing.

Cross-correlations were calculated for each pair of simultaneously recordedrespiratory and LFP signals, as well as the autocorrelation of each of the signals(Fig. 1). The respiratory frequency was identified as the peak frequency of thepower spectrum calculated from the respiratory signal after a bandpass filteringbetween 0.5 and 10 Hz (the 4th order Butterworth filter).

Phase locking between the respiratory rhythm or nasal airflow and the LFPoscillations was assessed by computing the coherence between the two signals. Thecoherency spectrum was estimated in the following manner: 50-s segments ofresting respiration and LFP signals served for the estimation of respiration-LFPcoherence. Each of these segments were further segmented in 19 subsegments of 5 sduration with 2.5 s overlap. For each of those subsegments, we calculated the





autospectrum separately of each signals and the cross-spectra of the respiration-LFP signals. The auto- and cross-spectra were averaged over the 50-s segment, andthe average cross-spectrum was normalized by the square roots of the average auto-spectra of the respiration and LFP signals. To compare the degrees of therespiration-LFP coherence in normal and bulbectomized mice, we took from eachcoherency spectrum the coherence value at the corresponding respirationfrequency and tested for the significance of the difference in the coherence valuesfor the two groups with the Mann–Whitney U-test. When we performed thecoherence analysis in a time-resolved manner, we sampled consecutive 10-ssegments of 9-s overlaps (that is, shifts in steps of 1 s) and applied the above-mentioned coherence spectrum estimation to each of the 10-s segments with 1-ssubsegments of 0.5-s overlaps.

To study cross-frequency phase–amplitude coupling of the LFP activity, weextracted from the LFP recordings the oscillatory phase ffp

ðtÞ and amplitude Afa ðtÞat time t and at two different frequencies, respectively, that is, phase frequency fp

and amplitude frequency fa. For this extraction, we used the wavelet transform witha Morlet wavelet defined at time t and frequency f as

ffiffiffif

p� exp½i2pf ðu� tÞ� exp

�ðu� tÞ2= 2s2ð Þ� �

. The parameter s was set to 10/(6f) so that the waveletcontains about 10 cycles of oscillations. The obtained amplitude was divided by itsmean so that the absolute amplitude value, which depends on the frequency, doesnot affect the results of the forthcoming phase–amplitude coupling analysis.

The degree of the phase–amplitude coupling between the frequencies fa and fp

was measured with the mean vector length MVL ¼ 1T

R T0 Xfa ;fp ðtÞdt

�� (Equation (1)),

where T is the duration of the signal and Xfa,fpis the composite signal defined as

Xfa ;fp ðtÞ � Afa ðtÞ exp iffpðtÞ

h i(Equation (2)). The phase frequency fp was fixed to

the breathing frequency, estimated as the peak frequency of the power spectrum ofthe respiratory signal. The amplitude frequency fa was varied in a range offrequencies between 1 and 256 Hz, sampled evenly on a logarithmic scale. To testthe significance of the mean vector length, 1,000 phase-randomized surrogateswere generated from the original signal by applying random phase shift to everyFourier coefficient of the original signal in the frequency domain and then inverse-Fourier transforming the randomized signal back into the time domain. The meanvector length was computed for each of them in exactly the same manner as for theoriginal signal. The original mean vector length was considered to be significantlyhigh if it exceeded the 95 percentile of the surrogate mean vector lengths.

The mean vector length was computed for 50-s segments of the recordingssampled from artefact-free periods. For each of the normal mice, 2–3 samples ofsuch artefact-free segments were obtained. For each of the bulbectomized mice, onesegment of 50-s was taken.

Analysis of spiking activity. In two awake, head-fixed mice, we recorded spikingactivity in the whisker barrel cortices while also monitoring respiratory behaviour.Spiking activity was recorded at a total of 12 sites, six in each mouse. Spike eventswere extracted using a fixed threshold. If the signal to noise ratio of a spike signalwas larger than four and the inter-spike intervals had a gamma distribution with arefractory period 45 ms we regarded it as a single unit (n¼ 3). All other spiketrains were regarded as multi-unit signals (n¼ 9). The spike trains were convolvedwith a Gaussian of unit area and s¼ 2.5 ms, and the resulting function was cor-related with the respiratory signal (that is, the thermistor voltage output). Todetermine the significance of spike-respiration correlations, we generated surrogatecorrelations by randomly shifting the spike train in time relative to the respiratorytimes. The spike train could be shifted in time by any random value between 0 sand the full duration of the recording. All spike times were shifted by the samevalue, preserving the spike interval sequence. The process was repeated 100 times,each time calculating the correlation of the shifted spike trains and saving thecorrelation values. From these 100 surrogate correlations, we calculated the medianand the 5th and 95th percentile of the distributions of correlation coefficients ateach lag time. Correlations were considered significant if the raw correlationcoefficient values exceeded the 5th or 95th percentile of the surrogate distributions(Fig. 2).

Tracing studies. Five mice were anaesthetized (i.p. injection) with Ketamine/Xylazine (100/10 ml kg� 1) and positioned in a stereotaxic frame (Stoelting, WoodDale, IL, USA). The scalp was opened with a midline incision, and the skull waslevelled between bregma and lambda by adjusting the bite bar. Body temperaturewas maintained at 35 �C using a heating pad. A glass micropipette filled with 5%Fluorogold (FG; Fluorochrome, LLC, Denver, CO, USA) was lowered into barrelcortex (anteroposterior¼ � 1.0 mm, mediolateral¼ 3.8 mm, dorsoventral¼ � 0.3mm), and the tracer was injected via iontophoresis (Precision Current Source,Stoelting Co., Wood Dale, IL, USA), at 2 mA (cycle 8 s ON and 8 s OFF), for a totalof 20 min. The injection pipette was left in place for 10 min before and after theinjection. Supplemental anaesthetic was administered as necessary throughout thesurgery to maintain the animals under deep anaesthesia.

After a 5-day survival period, mice were perfused transcardially withphosphate-buffered saline and 4% paraformaldehyde. The brains were removedand placed in 4% paraformaldehyde for 1 day and then transferred to a 30%buffered sucrose solution and stored at 4 �C for at least 1 week. Coronal sections(40 mm, every other section) were cut serially using a freezing microtome. FG canbe visualized using fluorescent microscopy, so no further histological processing

was necessary. Sections were rinsed, mounted, air-dried and coverslipped on saline-coated slides with DPX mounting media for histology. All material were visualizedand imaged using a Leica (DMRXA2, Leica Microsystems, Bannockburn, IL, USA)episcopic-fluorescence microscope equipped with a digital camera (HamamatsuORCA-ER, Hamamatsu , Shizuoka, Japan) and imaging software (SimplePCI,Hamamatsu, Shizuoka, Japan) (Fig. 7).

References1. Buzsaki, G. Rhythms of the Brain (Oxford University Press, 2006).2. Buzsaki, G. & Draguhn, A. Neuronal oscillations in cortical networks. Science

304, 1926–1929 (2004).3. Buzsaki, G. & Wang, X. J. Mechanisms of gamma oscillations. Annu. Rev.

Neurosci. 35, 203–225 (2012).4. Amzica, F. & Steriade, M. Electrophysiological correlates of sleep delta waves.

Electroencephalogr. Clin. Neurophysiol. 107, 69–83 (1998).5. Timofeev, I. & Chauvette, S. Thalamocortical oscillations: local control of EEG

slow waves. Curr. Top. Med. Chem. 11, 2457–2471 (2011).6. Ros, H., Sachdev, R. N., Yu, Y., Sestan, N. & McCormick, D. A. Neocortical

networks entrain neuronal circuits in cerebellar cortex. J. Neurosci. 29,10309–10320 (2009).

7. Petersen, C. C., Hahn, T. T., Mehta, M., Grinvald, A. & Sakmann, B. Interactionof sensory responses with spontaneous depolarization in layer 2/3 barrel cortex.Proc. Natl Acad. Sci. USA 100, 13638–13643 (2003).

8. Poulet, J. F. & Petersen, C. C. Internal brain state regulates membranepotential synchrony in barrel cortex of behaving mice. Nature 454, 881–885(2008).

9. Cao, Y., Roy, S., Sachdev, R. N. & Heck, D. H. Dynamic correlationbetween whisking and breathing rhythms in mice. J. Neurosci. 32, 1653–1659(2012).

10. Moore, J. D. et al. Hierarchy of orofacial rhythms revealed through whiskingand breathing. Nature 497, 205–210 (2013).

11. Deschenes, M., Moore, J. & Kleinfeld, D. Sniffing and whisking in rodents.Curr. Opin. Neurobiol. 22, 243–250 (2012).

12. Ward, L. M. Synchronous neural oscillations and cognitive processes. TrendsCogn. Sci. 7, 553–559 (2003).

13. Fries, P. A mechanism for cognitive dynamics: neuronal communicationthrough neuronal coherence. Trends Cogn. Sci. 9, 474–480 (2005).

14. Canolty, R. T. & Knight, R. T. The functional role of cross-frequency coupling.Trends Cogn. Sci. 14, 506–515 (2010).

15. Ito, J., Maldonado, P. & Grun, S. Cross-frequency interaction of the eye-movement related LFP signals in V1 of freely viewing monkeys. Front Syst.Neurosci. 7, 1 (2013).

16. Adrian, E. D. The role of air movement in olfactory stimulation. J. Physiol. 114,4–5p (1951).

17. Phillips, M. E., Sachdev, R. N., Willhite, D. C. & Shepherd, G. M. Respirationdrives network activity and modulates synaptic and circuit processing of lateralinhibition in the olfactory bulb. J. Neurosci. 32, 85–98 (2012).

18. Fontanini, A. & Bower, J. M. Variable coupling between olfactory systemactivity and respiration in ketamine/xylazine anesthetized rats. J. Neurophysiol.93, 3573–3581 (2005).

19. Fontanini, A., Spano, P. & Bower, J. M. Ketamine-xylazine-induced slow(o1.5Hz) oscillations in the rat piriform (olfactory) cortex are functionallycorrelated with respiration. J. Neurosci. 23, 7993–8001 (2003).

20. Han, F., Subramanian, S., Price, E. R., Nadeau, J. & Strohl, K. P. Periodicbreathing in the mouse. J. Appl. Physiol. 92, 1133–1140 (2002).

21. Sun, H., Luhmann, H. J. & Kilb, W. Resonance properties of different neuronalpopulations in the immature mouse neocortex. Eur. J. Neurosci. 36, 2753–2762(2012).

22. Canolty, R. T., Cadieu, C. F., Koepsell, K., Knight, R. T. & Carmena, J. M.Multivariate phase-amplitude cross-frequency coupling in neurophysiologicalsignals. IEEE Trans. Biomed. Eng. 59, 8–11 (2012).

23. Canolty, R. T. et al. Oscillatory phase coupling coordinates anatomicallydispersed functional cell assemblies. Proc. Natl Acad. Sci. USA 107,17356–17361 (2010).

24. Masaoka, Y., Satoh, H., Akai, L. & Homma, I. Expiration: the moment weexperience retronasal olfaction in flavor. Neurosci. Lett. 473, 92–96 (2010).

25. Masaoka, Y., Yoshimura, N., Inoue, M., Kawamura, M. & Homma, I.Impairment of odor recognition in Parkinson’s disease caused by weakactivations of the orbitofrontal cortex. Neurosci. Lett. 412, 45–50 (2007).

26. Brown, C. E. & Dyck, R. H. Retrograde tracing of the subset of afferentconnections in mouse barrel cortex provided by zincergic neurons. J. Comp.Neurol. 486, 48–60 (2005).

27. Adrian, E. D. The electrical activity of the mammalian olfactory bulb.Electroencephalogr. Clin. Neurophysiol. 2, 377–388 (1950).

28. Grosmaitre, X., Santarelli, L. C., Tan, J., Luo, M. & Ma, M. Dual functions ofmammalian olfactory sensory neurons as odor detectors and mechanicalsensors. Nat. Neurosci. 10, 348–354 (2007).





29. Welker, W. I. Analysis of sniffing of the albino rat. Behaviour 22, 223–243(1964).

30. Ni, H. et al. Upregulation of barrel GABAergic neurons is associated with cross-modal plasticity in olfactory deficit. PLoS One 5, e13736 (2010).

31. Ye, B. et al. The functional upregulation of piriform cortex is associatedwith cross-modal plasticity in loss of whisker tactile inputs. PLoS One 7, e41986(2012).

32. Hamada, Y., Miyashita, E. & Tanaka, H. Gamma-band oscillations in the‘barrel cortex’ precede rat’s exploratory whisking. Neuroscience 88, 667–671(1999).

33. Jones, M. S. & Barth, D. S. Sensory-evoked high-frequency (gamma-band)oscillating potentials in somatosensory cortex of the unanesthetized rat. BrainRes. 768, 167–176 (1997).

34. Bryant, J. L., Roy, S. & Heck, D. H. A technique for stereotaxic recordings ofneuronal activity in awake, head-restrained mice. J. Neurosci. Methods 178, 75(2009).

35. Hayar, A., Bryant, J. L., Boughter, J. D. & Heck, D. H. A low-cost solution tomeasure mouse licking in an electrophysiological setup with a standard analog-to-digital converter. J. Neurosci. Methods 2, 203–207 (2006).

AcknowledgementsWe thank Shuhua Qi for techncal assistance and Michael Nguyen for custom-machinedparts. We also thank S. du Lac, D. Linden, M. Ennis and H. Kita for valuable suggestionsand comments on earlier versions on this manuscript. This work was supported bygrants NS060887, NS067201 from the National Institute of Health to D.H.H. and by theHelmholtz Alliance on Systems Biology, and Helmholtz Portfolio Supercomputing andModeling for the Human Brain (SMHB) and BrainScales (EU Grant 269912) to S.G.Additional intramural support to D.H.H. was obtained from the UTHSC College ofMedicine.

DisclaimerThe content of this publication is solely the responsibility of the authors and does notnecessarily represent the official views of the NIH.

Author contributionsD.H.H. conceived the original idea. D.H.H., S.R., Y.L., Y.C. and M.F. designed theexperiment. S.R., Y.L. and Y.C. performed the experiments. D.H.H. participated in partof the experiments. L.L and J.B. designed and performed the tracing study. J.I. performedthe majority of data analysis, S.G. advised the data analysis. D.H.H. contributed to dataanalysis. D.H.H., J.I. and S.G. wrote the paper. S.R., Y.L., J.B. and Y.C. contributed to thewriting.

Additional informationCompeting financial interests: The authors declare no competing financial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

How to cite this article: Ito, J. et al. Whisker barrel cortex delta oscillations andgamma power in the awake mouse are linked to respiration. Nat. Commun. 5:3572doi: 10.1038/ncomms4572 (2014).

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License. The images or

other third party material in this article are included in the article’s Creative Commonslicense, unless indicated otherwise in the credit line; if the material is not included underthe Creative Commons license, users will need to obtain permission from the licenseholder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-sa/3.0/




http://npg.nature.com/reprintsand

http://npg.nature.com/reprintsand

http://creativecommons.org/licenses/by-nc-sa/3.0/

http://creativecommons.org/licenses/by-nc-sa/3.0/


whisker barrel cortex delta oscillations and gamma power in the

Documents